On Composable and Parametric Uncertainty in Systems Co-Design
Yujun Huang, Marius Furter, Gioele Zardini · 2025
Read On Composable and Parametric Uncertainty in Systems Co-Design by Yujun Huang, Marius Furter, Gioele Zardini (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-02766.txt.
Introduction
Yujun Huang, Marius Furter, and Gioele Zardini extend a design theory in a paper from two thousand twenty-five. Co-design tackles complex systems where choices, parts, and goals are interdependent. Their monotone theory models systems as design problems over ordered sets of resources and functions.
Current tools handle uncertainty only as intervals, the best and worst case bounds. That misses risk and confidence, which many fields need. The authors give a unified framework that covers intervals, probability distributions, and parameterized models together. It supports queries about risk trade-offs, experiment design, and multi-stage decisions. A case study on task-driven drones shows the framework in use. The goal is co-design that can reason about uncertainty in a richer way.
Catalog record
| CREATOR | Yujun Huang, Marius Furter, Gioele Zardini |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 48 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-02766
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On Composable and Parametric Uncertainty in Systems Co-Design Yujun Huang, Marius Furter, Gioele Zardini Abstract Optimizing the design of complex systems requires navigating interdependent decisions, heterogeneous components, and multiple objectives. Our monotone theory of co-design offers a compositional framework for addressing this challenge, modeling systems as design problems (DPs) , representing trade-offs between functionalities and resources within partially ordered sets. While current approaches model uncertainty using intervals, capturing worst- and best-case bounds, they fail to express probabilistic notions such as risk and confidence. These limitations hinder the applicability of co-design in domains where uncertainty plays a critical role. In this paper, we introduce a unified framework for composable uncertainty in co-design, capturing intervals, distributions, and parametrized models. This extension enables reasoning about risk-performance trade-offs and supports advanced queries such as experiment design, learning, and multi-stage decision making. We demonstrate the expressiveness and utility of the framework via a numerical case study on the uncertainty-aware co-design of task-driven unmanned aerial vehicles (UAVs) . I Introduction Designing embodied systems involves complex trade-offs between hardware components, such as sensors, actuators, and processors, and software modules for perception, planning, and control [ 1 , 2 , 3 , 4 , 5 ] . Traditional methods often optimize subsystems in isolation, limiting both modularity and interdisciplinary collaboration [ 6 , 5 ] . Over the past few years, we have introduced a monotone framework for co-design, which allows one to formulate and solve complex, compositional design optimization problems leveraging domain theory and category theory [ 5 ] . The existing toolbox has been successfully applied to solve problems in robotics and controls [ 7 , 8 , 9 ] , transportation [ 10 ] , and automotive [ 11 ] . However, existing approaches model uncertainty only via interval bounds, which guarantee robustness but lack the expressiveness needed to capture risk, probability of success, or adaptive decision-making under uncertainty. In this work, we extend the co-design framework to handle richer forms of uncertainty, including distributions and parametrized models, enabling queries over probabilistic trade-offs and paving the way for learning, estimation, and adaptive optimization in the design of complex systems. Specifically, we formalize uncertainty as composable structures over DPs , preserving the compositionality of co-design operations. We illustrate our approach through the co-design of UAVs , showing how uncertainty-aware design unlocks new capabilities for robust and efficient compositional decision-making. Organization of the paper The remainder of this paper is organized as follows. Section II reviews the foundations of our monotone co-design theory, including the current approach to uncertainty quantification via intervals. Section III extends the existing theory to model distributional and parameteric uncertainty in DPs . Furthermore, Section IV presents a case study on UAV co-design to illustrate the concepts introduced, and Section V concludes the work. II Monotone Co-Design Theory After introducing the required preliminaries, we summarize the main concepts of monotone co-design [ 12 , 13 , 5 ] . DP resources functionalities (a) A DP is a monotone relation between posets of functionalities and resources . d e \preceq (b) Series. d e (c) Parallel. d \succeq (d) Loop. Fig. 1 : MDPIs can be composed in different ways. II-A Mathematic preliminaries II-A 1 Sets and functions We write f\colon A\to B for functions between sets A and B and indicate the action of f on elements by m_{A}\mapsto f(m_{A}) . We call A the domain of f , and B its co-domain . We will often use the broad term map to refer to functions. Given maps f\colon A\to B and g\colon B\to C , their composite is the map g\circ f\colon A\to C that sends m_{A}\mapsto g(f(m_{A})) . We will often express composition diagrammatically: A\overset{f}{\to}B\overset{g}{\to}C . We write A\times B for the Cartesian product of sets. Its elements are tuples \langle m_{A},m_{B}\rangle , where m_{A}\in A and m_{B}\in B . Given maps f\colon A\to B and g\colon A^{\prime}\to B^{\prime} , their product is \displaystyle f\times g\colon A\times A^{\prime} \displaystyle\to B\times B^{\prime}, \displaystyle\langle m_{A},m_{A}^{\prime}\rangle \displaystyle\mapsto\langle f(m_{A}),g(m_{A}^{\prime})\rangle. Given h\colon A\times B\to C , we denote its partial evaluation by \displaystyle h(-,m_{B})\colon A \displaystyle\to C, \displaystyle m_{A} \displaystyle\mapsto h(m_{A},m_{B}). II-A 2 Background on orders Definition 1 (Poset) . A partially ordered set (poset) is a tuple \mathcal{P}=\langle P,\preceq_{\mathcal{P}}\rangle , where P is a set and \preceq_{\mathcal{P}} is a partial order (a reflexive, transitive, and antisymmetric relation). If clear from context, we use P for a poset , and \preceq for its order. Definition 2 (Opposite poset) . The opposite of a poset \mathcal{P}=\langle P,\preceq_{\mathcal{P}}\rangle is the poset \mathcal{P}^{\mathrm{op}}\mathrel{\raisebox{-1.29167pt}{$\overset{\text{\tiny def}}{=}$}}\langle P,\preceq_{\mathcal{P}}^{\mathrm{op}}\rangle with the same elements and reversed ordering: x_{P}\preceq_{\mathcal{P}}^{\mathrm{op}}y_{P}\Leftrightarrow y_{P}\preceq_{\mathcal{P}}x_{P} . Definition 3 (Product poset) . Given posets \langle P,\preceq_{\mathcal{P}}\rangle and \langle Q,\preceq_{\mathcal{Q}}\rangle , their product \langle P\times Q,\preceq_{\mathcal{P}\times\mathcal{Q}}\rangle is the poset with \langle x_{P},x_{Q}\rangle\preceq_{\mathcal{P}\times\mathcal{Q}}\langle y_{P},y_{Q}\rangle\Leftrightarrow(x_{P}\preceq_{\mathcal{P}}y_{P})\wedge(x_{Q}\preceq_{\mathcal{Q}}y_{Q}). Definition 4 (Upper closure) . Let P be a poset . The upper closure of a subset X_{P}\subseteq P contains all elements of P that are greater or equal to some y_{P}\in X_{P} : \uparrow\!X_{P}\mathrel{\raisebox{-1.29167pt}{$\overset{\text{\tiny def}}{=}$}}\{x_{P}\in P\mid\exists y_{P}\in X_{P},y_{P}\preceq_{P}x_{P}\}. Definition 5 (Upper set) . A subset X_{P}\subseteq P of a poset is called an upper set if it is upwards closed: \uparrow\!X_{P}=X_{P} . We write \mathtt{U}(P) for the set of upper sets of P . We regard \mathtt{U}(P) as partially ordered under U\preceq U^{\prime}\Leftrightarrow U\supseteq U^{\prime} . Definition 6 (Monotone map) . A map f\colon P\to Q between posets \langle P,\preceq_{\mathcal{P}}\rangle and \langle Q,\preceq_{\mathcal{Q}}\rangle is monotone if x\preceq_{\mathcal{P}}y implies f(x)\preceq_{\mathcal{Q}}f(y) . Monotonicity is preserved by composition and products. II-A 3 Background on probability Definition 7 (Measurable spaces) . A sigma algebra \Sigma_{A} on set A is a non-empty collection of subsets of A that is closed under complements, countable unions, and countable intersections. The tuple \langle A,\Sigma_{A}\rangle is called a measurable space . For an arbitrary collection of subsets \mathcal{G} , the sigma algebra \sigma(\mathcal{G}) generated by \mathcal{G} is the smallest sigma algebra containing \mathcal{G} . Given another measurable space \langle B,\Sigma_{B}\rangle , a map f\colon A\to B is called measurable if pre-images of measurable sets are measurable: f^{-1}(Y)\in\Sigma_{A},\ \forall Y\in\Sigma_{B} . Definition 8 (Probability distribution) . A probability distribution on a measurable space \langle A,\Sigma_{A}\rangle is a map \mathbb{P}\colon\Sigma_{A}\to[0,1] satisfying \mathbb{P}(A)=1 and \mathbb{P}(\bigcup_{i=1}^{\infty}E_{i})=\sum_{i=1}^{\infty}\mathbb{P}(E_{i}) for any collection \{E_{i}\}_{i=1}^{\infty} of disjoint sets. Elements in A are called outcomes or results and sets in the sigma algebra are called events . \mathbb{P}(E) denotes the probability of event E . II-B Monotone co-design theory Co-design provides a compositional framework for the analysis of complex systems. Consider the design of autonomous vehicles, where perception is an important sub-system. With other conditions fixed, the perception sub-system can be viewed as providing a certain level of detection accuracy at the cost of computation power , under given weather conditions . While detection accuracy and computation power can be naturally modeled as positive real numbers, weather conditions are complex and involve non-comparable elements. For instance, a clear night and a foggy day pose qualitatively different challenges to the perception system. Hence, designs optimized for one case will not necessarily perform well under the other. Monotone co-design formalizes such situations as DPs which provide functionalities at the cost of resources , both assumed to be partially ordered. This enables expressing trade-offs between incomparable optimal designs. Definition 9